What You'll Need
A GPU with at least 8GB VRAM (RTX 3070 or better recommended)ComfyUI installed on your systemWAN 2.2 model downloaded
Step 1: Install ComfyUI
ComfyUI is a node-based interface for Stable Diffusion that makes it easy to create complex generation workflows.
Clone the ComfyUI repositoryInstall the required Python dependenciesDownload the WAN 2.2 model and place it in the models directory
Step 2: Set Up Your First Workflow
Start with a simple txt2img workflow:
Add a "Load Checkpoint" node and select WAN 2.2Add a "CLIP Text Encode" node for your promptConnect them to a "KSampler" nodeAdd a "VAE Decode" node and "Save Image" node
Step 3: Generate Your First Video
For video generation with WAN 2.2:
Set the frame count to 16-32 framesUse a resolution of 832x1216 for portrait or 1216x832 for landscapeRecommended settings: 30 steps, CFG 7.5, Euler sampler
Tips for Better Results
Use **-tune animation** flag when encoding with FFmpegAdd quality tags like "masterpiece, best quality" to your promptsExperiment with different LoRA models for style variation